—Depression is a large-scale mental health problem and a challenging area for machine learning researchers in detection of depression. Datasets such as Distress Analysis Interview Corpus-Wizard of Oz (DAIC-WOZ) have been created to aid research in this area. However, on top of the challenges inherent in accurately detecting depression, biases in datasets may result in skewed classification performance. In this paper we examine gender bias in the DAIC-WOZ dataset. We show that gender biases in DAIC-WOZ can lead to an overreporting of performance. By different concepts from Fair Machine Learning, such as data redistribution , and using raw audio features, we can mitigate against the harmful effects of bias.
CCD (charged coupled device) and CMOS imaging technologies can be applied to thin tissue autoradiography as potential imaging alternatives to using conventional film. In this work, we compare two particular devices: a CCD operating in slow scan mode and a CMOS-based active pixel sensor, operating at near video rates. Both imaging sensors have been operated at room temperature using direct irradiation with images produced from calibrated microscales and radiolabelled tissue samples. We also compare these digital image sensor technologies with the use of conventional film. We show comparative results obtained with (14)C calibrated microscales and (35)S radiolabelled tissue sections. We also present the first results of (3)H images produced under direct irradiation of a CCD sensor operating at room temperature. Compared to film, silicon-based imaging technologies exhibit enhanced sensitivity, dynamic range and linearity.
The availability of audio data on sound sharing platforms such as Freesound gives users access to large amounts of annotated audio. Utilising such data for training is becoming increasingly popular, but the problem of label noise that is often prevalent in such datasets requires further investigation. This paper introduces ARCA23K, an Automatically Retrieved and Curated Audio dataset comprised of over 23 000 labelled Freesound clips. Unlike past datasets such as FSDKaggle2018 and FSDnoisy18K, ARCA23K facilitates the study of label noise in a more controlled manner. We describe the entire process of creating the dataset such that it is fully reproducible, meaning researchers can extend our work with little effort. We show that the majority of labelling errors in ARCA23K are due to out-of-vocabulary audio clips, and we refer to this type of label noise as open-set label noise. Experiments are carried out in which we study the impact of label noise in terms of classification performance and representation learning.
Basidiomycete fungi are a rich source of natural products with a diverse array of potentially exploitable bioactivities. Two dimeric sesquiterpenes, bovistol B (1) and D (2), and one monomeric sesquiterpene, strossmayerin (7), were isolated from the culture filtrate of the basidiomycete fungus Coprinopsis strossmayeri. The structures were determined through a combination of MS and 1D/2D NMR spectroscopic techniques. Likely monomeric precursors, identified on the basis of HRMS analysis, allow a plausible biosynthetic pathway to be proposed for the biosynthesis of 1 and 2, involving the dimerisation of the monomer through a hetero-Diels-Alder mechanism. A gene cluster, including a putative sesquiterpene 1–11 cyclase, was identified through phylogenetic and RNA-seq analysis, and is proposed to be responsible for the biosynthesis of 1 and 2.
Autoradiography is a widely extended pre-clinical nuclear imaging modality used in life sciences to investigate and localise radiolabelled biological pathways in thin ex-vivo tissue sections. After the tissue section has been exposed to an ionising radiation detector the resulting labelled regions are subsequently analysed. Typically, the resulting autoradiograms are analysed manually by an expert life scientists using a visual template as reference to measure the different radioligand uptake levels in the different areas of, in our case, mouse brain. This process is extremely time consuming and error prone, with the expertise of the life scientist playing a significant role. In this paper we describe a semi-automatic method to register a template brain atlas on to the brain autoradiogram making the analysis process more efficient, repeatable and independent of the expertise of the life scientist. The method first identifies those regions with high and low level of radioligand uptake by region growing segmentation. Subsequently, the counterpart regions in the corresponding atlas image are manually identified. Finally a set of control points is extracted from each region contour in the autoradiogram and the atlas image to apply a scattered data interpolator. ©2009 IEEE.
Segmentation in medical imaging plays a critical role easing the delineation of key anatomical functional structures in all the imaging modalities. However, many segmentation approaches are optimized with the assumption of high contrast, and then fail when segmenting poor contrast to noise objects. The number of approaches published in the literature falls dramatically when functional imaging is the aim. In this paper a feature extraction based approach, based on region growing, is presented as a segmentation technique suitable for poor quality (low Contrast to Noise Ratio CNR) images, as often found in functional images derived from Autoradiography. The region growing combines some modifications from the typical region growing method, to make the algorithm more robust and more reliable. Finally the algorithm is validated using synthetic images and biological imagery.
Autoradiography is a well established imaging modality in Biology and Medicine. This aims to measure the location and concentration of labelled molecules within thin tissue sections. The brain is the most anatomically complex organ and identification of neuroanatomical structures is still a challenge particularly when small animals are used for pre-clinical trials. High spatial resolution and high sensitivity are therefore necessary. This work shows the performance and ability of a prototype commercial system, based on a Charged-Couple Device (CCD), to accurately obtain detailed functional information in brain Autoradiography. The sample is placed in contact with the detector enabling direct detection of β- particles in silicon, and the system is run in a range of quasi-room temperatures (17-22 °C) under stable conditions by using a precision temperature controller. Direct detection of β- particles with low energy down to ~5 keV from 3[H] is possible using this room temperature approach. The CCD used in this work is an E2V CCD47-20 frame-transfer device which removes the image smear arising in conventional full-frame imaging devices. The temporal stability of the system has been analyzed by exposing a set of 14[C] calibrated microscales for different periods of time, and measuring the stability of the resultant sensitivity and background noise. The thermal performance of the system has also been analyzed in order to demonstrate its capability of working in other life science applications, where higher working temperatures are required. Once the performance of the system was studied, a set of experiments with biological samples, labelled with typical β- radioisotopes, such as 3[H], has been carried out to demonstrate its application in life sciences.